Old CRC32Fingerprint() Versus Fingerprint()

[Karbon]

So I have some legacy code where some checksums were created using this code ( I think in PB 4.4).

(New Code)

Code: Select all

Str(Val("$"+Fingerprint(@temp_kb7,StringByteLength(temp_kb7,#PB_Ascii),#PB_Cipher_CRC32)))

(Old Code)

Code: Select all

Str(CRC32Fingerprint(@temp_kb7,StringByteLength(temp_kb7)))

Given the same data, the two checksums are different. I need to figure out how to make the first example generate the exact same checksum as the second (given the same data of course).

I don't have access to the manual from PB 4.40 right now to look and see if it is documented differently. Can anyone help me out?

[JHPJHP]

Hi Karbon,

Try the following:

Code: Select all

nLength = StringByteLength(temp_kb7, #PB_Ascii)
*temp_kb7 = AllocateMemory(nLength)
PokeS(*temp_kb7, temp_kb7, nLength, #PB_Ascii)
Fingerprint(*temp_kb7, nLength, #PB_Cipher_CRC32)

[Karbon]

Unfortunately that wasn't it (It's an all-ASCII program and I'm using the LTS version for exactly that reason).

Since the original function returned an integer and that's what was stored, and the new version returns a hex string, I'm actually comparing like this :

Code: Select all

Str(Val("$"+Fingerprint(@temp_kb7,StringByteLength(temp_kb7,#PB_Ascii),#PB_Cipher_CRC32)))

But still a no-go. I must be missing something since if it's a true CRC32 implementation the checksums shouldn't ever be different. I'm going to grab the old version and install it when I get back to the house, I just can't download it here over this crappy 4G connection.

[Karbon]

You're right, I only saw the original post. Same thing, though, you just unrolled it a bit in your last as far as I can tell..

Other than the #PB_ASCII I don't see any difference in your example and mine.. Am I missing something there?

[JHPJHP]

Hi Karbon,

Are you able to post the original string and the expected checksum?

[nco2k]

pb doesnt have unsigned longs and Str() expects a quad / Val() returns a quad.

old:

Code: Select all

MyLong.l = CRC32Fingerprint(?MyString, ?MyStringEnd - ?MyString)
Debug Str(MyLong)
Debug Str(CRC32Fingerprint(?MyString, ?MyStringEnd - ?MyString) & $FFFFFFFF)
Debug StrU(CRC32Fingerprint(?MyString, ?MyStringEnd - ?MyString), #PB_Long)
Debug UCase(RSet(Hex(CRC32Fingerprint(?MyString, ?MyStringEnd - ?MyString), #PB_Long), 8, "0"))

DataSection
  MyString:
  Data.a $48, $65, $6C, $6C, $6F
  MyStringEnd:
EndDataSection

new:

Code: Select all

UseCRC32Fingerprint()

MyLong.l = Val("$"+Fingerprint(?MyString, ?MyStringEnd - ?MyString, #PB_Cipher_CRC32))
Debug Str(MyLong)
Debug Str(Val("$"+Fingerprint(?MyString, ?MyStringEnd - ?MyString, #PB_Cipher_CRC32)))
Debug UCase(RSet(Fingerprint(?MyString, ?MyStringEnd - ?MyString, #PB_Cipher_CRC32), 8, "0"))

DataSection
  MyString:
  Data.a $48, $65, $6C, $6C, $6F
  MyStringEnd:
EndDataSection

i still hope that regular Str() gets an optional type like StrU() - http://www.purebasic.fr/english/viewtop ... =3&t=64823

c ya,
nco2k

[Karbon]

So shouldn't these be producing the same string?

New

Code: Select all

Str(Val("$"+Fingerprint(@temp_kb7,StringByteLength(temp_kb7,#PB_Ascii),#PB_Cipher_CRC32)))

Old

Code: Select all

Str(CRC32Fingerprint(@temp_kb7,StringByteLength(temp_kb7)))

[nco2k]

no, because Val("$"+X) will return the unsigned value of the hex string.

Code: Select all

MyQuad.q = Val("$"+Fingerprint(@temp_kb7,StringByteLength(temp_kb7,#PB_Ascii),#PB_Cipher_CRC32))
Debug Str(MyQuad)
MyLong.l = Val("$"+Fingerprint(@temp_kb7,StringByteLength(temp_kb7,#PB_Ascii),#PB_Cipher_CRC32))
Debug Str(MyLong)

c ya,
nco2k

[Karbon]

Well, damn... I thought you were saying your old/new examples above were equivalents.

Do you have any idea what the old CRC32Fingerprint() returned?

[nco2k]

Well, damn... I thought you were saying your old/new examples above were equivalents.

they are. just run them when u get home.

Do you have any idea what the old CRC32Fingerprint() returned?

yes, a signed long.

c ya,
nco2k

[Karbon]

Awesome -- thank you!